Download Aalborg Universitet 5G small cell optimized radio design

Aalborg Universitet
5G small cell optimized radio design
Mogensen, Preben Elgaard; Pajukoski, Kari; Tiirola, Esa ; Lähetkangas, Eeva; Vihriälä,
Jaakko; Vesterinen, Seppo; Laitila, Matti; Berardinelli, Gilberto; Da Costa, Gustavo Wagner
Oliveira; Garcia, Luis Guilherme Uzeda; Tavares, Fernando Menezes Leitão; Cattoni, Andrea
Fabio
Published in:
Globecom. I E E E Conference and Exhibition
DOI (link to publication from Publisher):
10.1109/GLOCOMW.2013.6824971
Publication date:
2013
Document Version
Early version, also known as pre-print
Link to publication from Aalborg University
Citation for published version (APA):
Mogensen, P., Pajukoski, K., Tiirola, E., Lähetkangas, E., Vihriälä, J., Vesterinen, S., ... Cattoni, A. F. (2013). 5G
small cell optimized radio design. In Globecom. I E E E Conference and Exhibition. (pp. 111-116). IEEE.
(Globecom. I E E E Conference and Exhibition). DOI: 10.1109/GLOCOMW.2013.6824971
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
? Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
? You may not further distribute the material or use it for any profit-making activity or commercial gain
? You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us at [email protected] providing details, and we will remove access to
the work immediately and investigate your claim.
Downloaded from vbn.aau.dk on: September 17, 2016
5G small cell optimized radio design
Preben Mogensen (1, 2), Kari Pajukoski (3), Esa Tiirola (3), Eeva Lähetkangas (3), Jaakko Vihriälä (3),
Seppo Vesterinen (3), Matti Laitila (3), Gilberto Berardinelli (2), Gustavo W. O. Da Costa (2), Luis G. U. Garcia (2),
Fernando M. L. Tavares (2), Andrea F. Cattoni (2)
(1) Nokia Siemens Networks, Aalborg, Denmark
(2) Department of Electronic Systems, Aalborg University, Denmark
(3) Nokia Siemens Networks, Oulu, Finland
Abstract— The 5th generation (5G) of mobile radio access
technologies is expected to become available for commercial
launch around 2020. In this paper, we present our envisioned 5G
system design optimized for small cell deployment taking a clean
slate approach, i.e. removing most compatibility constraints with
the previous generations of mobile radio access technologies. This
paper mainly covers the physical layer aspects of the 5G concept
design.
I.
INTRODUCTION
Historically, new generations of radio access technologies
(RATs) have been introduced with an interval of approximately
ten years to cope with the exponential increase of the mobile
data traffic and to take full advantage of the evolution of the
technology components without any legacy burden [1].
The specifications of the Long Term Evolution - Advanced
(LTE-A) standard, which is agreed to be the 4th generation
(4G) mobile communication technology, were finalized back in
2010 [2]. If history is any indication, a new 5th Generation (5G)
radio standard is expected to reach the mass market around
2020 and to last until ~2030, where we may potentially
experience a 6th Generation (6G).
In [3] we have predicted the mobile traffic growth to be in
the range of ~x150-500 by 2020 (with reference to 2010), and
to increase to ~x3000-30,000 by 2030. We also predict the user
data rate demand to grow by a factor of ~x10 by 2020 and a
factor ~x100 by 2030. Such growth values in both traffic
volume and user data rate set demands to both capacity and
coverage of 5G.
Within the industry, it is generally anticipated that the 5G
network evolution will be heterogeneous in cell types - ranging
from macro to pico - and it will also integrate multiple radio
access technologies. Our ~x1000 HetNet evolution studies for
2020 indicate that both indoor and outdoor hot spot small cells
will start to play a very dominant role to meet the above
mentioned capacity and user data rate demands [3]. The studies
also clarify the need to operate such ultra dense deployment of
small cells in a dedicated spectrum to avoid coexistence issues
with high power micro/macro cells. The frequency band from
3.4-4.9 GHz has drawn attention for increasing capacity of
International Mobile Telecommunications (IMT) systems in
the short term (i.e., WRC-2015 5D agenda point) and
potentially even higher frequency bands to be allocated at the
WRC-2018/19. The usage of millimiter waves has also drawn
attention given the large amount of available bandwidth [4].
Bearing in mind the dedicated spectrum, our 5G studies
have focused on a clean slate approach for the design of a
Authors’ email addresses: {preben.mogensen, kari.pajukoski, esa.tiirola,
eeva.lahetkangas, jaakko.vihriala, seppo.vesterinen, matti.laitila}@nsn.com,
{gb, gc, lug, ft, afc}@es.aau.dk
novel RAT optimized for small cells. Given the historical
evolution of the technology components [5], the classical key
performance indicators of 5G are estimated to be significantly
better than 4G, i.e.
x peak data rates should be in the order of 10 Gbps;
x Round Trip Time (RTT) should be in the order of
1ms;
x spectral efficiency to be at least ~x2 better than 4G.
However, we also see new significant drivers for the 5G RAT
design, such as:
x Very low power consumption of both access points
and terminals
x Efficient support of Machine Type Communication
(MTC)
x Flexibility in spectrum usage
x Self-optimized ultra-dense deployment of access
points
x Support of multi-hop (e.g., self-backhauling)
x Simple and low cost design.
While our previous contribution [3] was focused on the
motivation for initiating Beyond 4G research and on a highlevel description of its technology enablers, this paper presents
our latest updates in the 5G small cell concept design.
The paper is structured as follows. The main criteria for the
physical layer design are described in Section II along with the
proposed frame structure and the envisioned numerology.
Section III focuses on the Radio Resource Management (RRM)
issues for interference mitigation, while Section IV discusses
the feasibility of network synchronization which represents an
underlying assumption for our design. Section V presents a
general view on the 5G networking aspects. Finally, Section VI
resumes the conclusions and states the future work.
II.
PHYSICAL LAYER DESIGN
Our previous contribution [3] presented and motivated the
main physical layer technology components of our novel
Beyond 4G /5G RAT. It is agreed that a set of advanced
features such as Multiple-Input-Multiple-Output (MIMO)
antenna technology [6], fast link adaptation, Hybrid-Automatic
Repeat Request (HARQ) [7] and interference mitigation
techniques are to be included in the design. Orthogonal
Frequency Division Multiplexing (OFDM) is recognized as the
preferred modulation for both uplink and downlink given its
multipath mitigation capability and the straightforward
extension to MIMO [7]. Moreover, Time Division Duplex
(TDD) mode has been preferred to Frequency Division Duplex
Figure 1. 5G frame structure
(FDD) due to its cost-effectiveness as well as the possibility of
exploiting large unpaired frequency bands. The usage of TDD
mode also allows exploiting the channel reciprocity between
uplink (UL) and downlink (DL) for reducing the feedback
overhead; a preliminary over-the-air calibration procedure is
then necessary for aligning the radio chains of both Access
Point (AP) and User Equipment (UE).
The underlying assumption for our design is a fully
synchronized system, i.e. all APs/UEs in the network share the
same knowledge of the frame timing.
The aforementioned technology components are already well
established in the engineering community, and a more detailed
description of their usage has been presented in [3]. Here, we
focus on the design of the 5G frame structure and numerology.
II.A 5G Frame structure
The 5G physical layer aims at providing high performance in
terms of data rate and latency with reduced cost and power
consumption. A proper design of the frame structure is then
fundamental for achieving our ambitious targets. More
specifically, the frame should be designed bearing in mind the
following requirements:
x Low latency, below 1 ms. Such requirement
establishes a reduction of a factor of ~x10 with
respect to the LTE target.
x Support for frequency coordination and reuse. This is
meant for dealing with the co-channel interference
which may be a strong limiting factor in the 5G
performance due to the uncoordinated cell
deployment.
x Support for advanced receivers. The usage of such
receivers represents a further promising option for
counteracting the co-channel interference and then
boosting the data rate.
x Support for pipeline-processing at the receiver. This
reasonably leads to low computational complexity and
reduced latency.
x Low power consumption, allowing longer UE battery
life.
x Support for communication links beyond the
traditional AP-UE access, e.g., MTC and selfbackhauling.
The envisioned frame structure is shown in Figure 1. Note that
a short guard period (GP) is inserted at each potential switch of
the communication link, in order to accommodate the on-off
power transient.
In the frequency domain, the system bandwidth is divided into
a number of Physical Resource Blocks (PRBs) providing native
support for frequency coordination/reuse techniques. The first
part of the frame represents the control part while the data part
is located next.
In LTE, both Physical Downlink Control Channel (PDCCH)
and Physical Uplink Control Channel (PUCCH) are mapped
over a set of subcarriers in multiple OFDM symbols which also
carry data information [7]; here, the time separation between
control part and data part in the frame allows instead separating
control and data planes. This enables the cost-effective pipeline
processing at the receiver, since the UE can process its
dedicated control information while transmitting/receiving in
the data part, thus reducing the latency.
In case of traditional AP-UE link, the control part can be
composed for example of one symbol for the DL and one
symbol for the UL, as shown in Figure 1. Further, the control
symbols can be selected in a link centric manner to enable
further communication directions. For instance, the two
symbols can be acceded by devices communicating directly
with each other in case of MTC, or by the AP and a relay node
for self-backhauling.
The time separation between control and data part is also meant
to reduce the power consumption at the UE; the device can
turn-off its receiver chain for the rest of the frame in case it
does not receive any command or information in the control
part, thus reducing its battery consumption.
Examples of control information to be mapped in the control
symbols are the scheduling request in the UL, the scheduling
grant in the DL, the Modulation and Coding Scheme (MCS)
indicator for the Adaptive Modulation and Coding (AMC), the
Rank Indicator (RI) which sets the number of streams to be
used, etc. Such information can be represented by a small
number of bits which can be encoded with a fixed robust
modulation (e.g. QPSK 1/6).
Since the system is fully synchronized, multiple APs/UEs may
transmit control signaling employing the same PRBs
simultaneously, thus leading to harmful collisions; this is
precisely where Radio Resource Management (RRM)
techniques discussed in Section III come into the picture to
ensure a sensible selection of different PRBs. The subsequent
data part can be dedicated entirely to either UL transmission or
DL transmission. Its first symbol is dedicated to the
DeModulation Reference Symbols (DMRS), which are used
for channel estimation purposes. The well-known Zadoff-chu
sequences [7] can be used as DMRS due to their favorable
cross-correlation properties. This design allows the
simultaneous estimation of the channel responses of multiple
interfering APs/UEs in the data part. Moreover, it also
stabilizes the interference pattern within a radio frame, thus
being a pivotal enabler of the effective usage of interference
rejection combining (IRC) receivers [8], as will be further
discussed in Section III.
The ambitious latency target can be achieved by assuming a
frame duration of 0.25 ms, and an optimized scheduling and
HARQ design. Both scheduling and HARQ processes are
shown in Figure 2(a) and Figure 2(b), respectively.
The UE initiated data transmission requires 3 TDD cycles
(scheduling request in the UL, scheduling grant in the DL, data
transmission in the UL), for a total of 0.75 ms.
Similarly, the round trip time of the HARQ process (AP grant,
AP transmission and UE ACK/NACK transmission) requires
Table 1. 5G, LTE/LTE-A and IEEE 802.11ac numerologies
Figure 2. RTT for scheduling (a) and HARQ (b).
0.75 ms including processing times. Differently from LTETDD [7], the HARQ round trip time is here fixed and does not
depend on the UL/DL ratio; the control part in each radio
frame offers at least one OFDM symbol in each direction for
the transmission of acknowledgments. The number of parallel
HARQ processes is 4, while in LTE-TDD it is up to 15; this
allows a considerable reduction of the memory circuitry
(buffers), which leads to significant cost savings in the
baseband chip.
II.B 5G numerology
The envisioned 5G frame numerology is shown in Table 1.
LTE, LTE-A and IEEE 802.11 [9] numerologies are also
included for the sake of comparison. We assume as a baseline
for our initial studies a carrier bandwidth of 200 MHz, though
multiple carriers will be required for achieving the maximum
throughput target when combined with 4x4 MIMO and
256QAM modulation.
5G subcarrier spacing is set to be 4 times larger than the
LTE/LTE-A one, leading to a 4 times shorter time symbol
duration. Such large subcarrier spacing allows higher
robustness than LTE to the phase noise; while primarily
targeting frequency band from 3.4-4.9 GHz, 5G systems can be
then set to operate at significantly high carrier frequencies (e.g.
15 GHz) even with relatively cheap devices. On the other side,
the 5G symbol duration is still much larger than the IEEE
802.11 one. This allows maintaining a relatively low number of
symbols per frame, with the advantage of saving the
cumulative overhead given by the Cyclic Prefix (CP) at the
beginning of each symbol.
The CP duration is considerably lower than the LTE one, given
the shorter expected delay spread/propagation delay in local
area scenarios.
In LTE the minimum GP duration (in case no timing advance
procedure is taking place) is set to the OFDM symbol duration
in order to maintain the same numerology on both the data
frame and the special subframe where the switching is
operated, as well as for compatibility with existing standards
operating in the same bandwidth. Our clean slate design
approach for 5G removes any backwards compatibility
constraint; moreover, the transmit power of the AP in local area
is significantly lower than the micro and macro LTE base
stations. This leads to the possibility of setting an extremely
short GP for the on/off power transient.
Carrier Bandwidth
[MHz]
Subcarrier spacing
[kHz]
Symbol length [µs]
FFT size
Effective
subcarriers
TTI duration [ms]
Number of GPs
Number of
symbols per frame
CP duration [µs]
5G
LTE
LTE-A
(5 CCs)
200
20
100
20
160
60
15
15
312.5
312.5
16,67
4096
3300
66.67
2048
1200
66.67
5x2048
6000
4
64
56
4
512
484
0.25
3
14
1
2
14
1
2
14
variable
variable
none
n.a.
none
n.a.
1
4.7
(short)
66.67
(min)
7.25
4.7
(short)
66.67
(min)
7.25
0.4
(short)
none
0.4
(short)
none
11
11
up to
15
up to
75
none
none
GP duration [µs]
0.89
Overhead
(CP+GP) [%]
HARQ processes
6.67
4
802.11ac
Note that in 5G the resultant cumulative overhead represented
by CP and GP is approximately the same of LTE and LTE-A,
with the advantage of higher robustness to phase noise.
III.
RRM AND INTERFERENCE MANAGEMENT
Radio Resource Management (RRM) refers to a variety of
procedures with the specific aim of using the radio resources in
the most efficient way in terms of target metrics such as
throughput, reliability and latency. Broadly, a RRM framework
is taking care of performing functions such as:
x Selection of transmission modes and corresponding
parameters, e.g. power or Modulation and Coding
Scheme (MCS).
x Decision of which nodes will be transmitting,
receiving or idle in a particular radio resource
(frequency, time, spatial stream or code).
x Channel assignment for particular links and traffic
flows.
The possibility of separating control and data planes enabled
by the frame structure described in Section II also allows the
usage of separate RRM strategies for data and control, which
have very different requirements.
A correct reception of the control information is indeed critical
in any RAT for enabling data transmission. Control channel
reliability is always a design goal, even at the cost of a
significantly larger overhead, e.g. the usage of a fixed robust
modulation and coding scheme. Frequency domain Inter-cell
Interference Coordination (ICIC) techniques are yet another
safe-guard to ensure high Signal-to-Interference-plus-Noise
Ratio (SINR) for control channels. The selection of a subset of
PRBs for control channel operation can be done by efficient
distributed techniques at the time of bootstrapping or when
more control channel capacity is needed, using for example the
mechanism in [10].
A new and different RRM challenge lies in the foreseen usage
of cooperative link adaptation techniques or inter-cell
interference canceling [11]. In order to work properly, such
1
0.9
0.8
0.7
CDF
0.6
0.5
0.4
0.3
0.2
MRC R1
IRC R1
MRC R2
IRC R2
0.1
0
0
500
1000 1500
2000 2500 3000
Data Transmission Rate [Mbps]
3500
4000
Figure 3. Potential of IRC receiver in a local area 3GPP scenario.
schemes require that devices are able to decode the control
information from neighboring cells. The synchronized
transmission of control information in different PRBs is a first
step towards this goal. It enables the receiving APs/UEs to
know precisely when to attempt decoding such control
information (e.g., scheduling assignment) from neighboring
cells. Nevertheless, selecting the right set of PRBs that ensures
a reasonable SINR for both intended and neighboring receivers
is a topic for further studies.
On the data plane, it is typically worth to be more lax about
the interference effects. HARQ works indeed as an insurance
against exceptionally congested or faded frames. Moreover,
the fast varying nature of traffic will make very persistent
interference less likely. In this way, on the data plane one can
go after higher risks of losing a single data transmission with
higher payoffs in terms of throughput. Compared to the
control plane, the data plane has then an almost diametrically
opposed RRM strategy: heavier reliance on advanced
receivers, link and rank adaptation with a reactive application
of resource-limiting ICIC techniques as a self-healing
mechanism.
As mentioned in Section II, the interference is stable on a
frame basis even though the preferred direction of
transmission (UL or DL) is expected to vary quickly. Simply
put, the APs/UEs that are sending their reference sequences in
the DMRS symbol will also be the ones transmitting data in
the rest of the radio frame. This is particularly suited for the
usage of advanced receivers, e.g. IRC. Such receivers suppress
the interference by exploiting a periodical estimate of the
interference-plus-noise covariance matrix [8]. Given our frame
structure, such estimate can be performed from the DMRS
symbol at every frame, and then used for the computation of
the combining matrix that will maximize the SINR in the data
part. This design works regardless of the interference source,
i.e. the receiver is able to estimate the cross-link interference
(e.g., UE-to-UE) as well, solving one of the main concerns
related to uncoordinated TDD systems. In a previous work
[12], we have studied the potential benefits of the use of IRC
receivers in 3GPP-inspired local area scenarios with 40 cells
[13]. We compared the performance of IRC receivers to the
baseline Maximum Ratio Combining (MRC) receiver [14],
which does not exploit any interference estimate and
represents the baseline detector, for instance, in LTE. Figure 3
shows the Cumulative Distribution Function (CDF) of the
downlink data rates in a full buffer traffic scenario assuming
ideal link and rank adaptation with up to 4 MIMO spatial
streams. Both cases of frequency reuse 1 (R1) and planned
frequency reuse 2 (R2) are displayed. The results show that
IRC can significantly improve the data rates (especially for the
cells in the worst interference conditions) without bandwidth
sacrifices even in very demanding scenarios. Further details
are in [12].
Besides the advanced physical layer capabilities, our
envisioned 5G system has some additional peculiarities that
need to be factored into the RRM design for the 5G data
plane:
x multi-Gbps capabilities, which indirectly can lead to
wide traffic fluctuations as explained in more detail
below;
x fast variability of interference sources, due to
independent switching points per cell and
topological variations (MTC, self-backhauling).
From the traffic perspective, keeping a flow steadily
transmitting at multi-Gbps data rate is a very significant
challenge [15]. Networking buffers as well as flow and
congestion control mechanisms will tend to make the
instantaneous data rate vary quickly from very high to nearly
zero. The level of flow aggregation is rather small in a 5G cell
due to the presence of few users in local area, leading to an
unprecedented burstiness level. The ability to change the
transmission direction of the data part every 0.25 ms subframe
allows fast adaptation to the traffic demand. The system may
more freely schedule the oldest packet in uplink or downlink
queues. In LTE-A this is not possible, since the UL/DL
configuration determines the direction of each subframe.
Furthermore, in LTE-A at most 60% of resources can be
allocated to UL [2]. This presents a serious disadvantage in
UL-heavy traffic.
However, this flexibility comes at a price: rapid and
independent variations of transmission direction pose
significant challenges in terms of link/rank adaptation, since
the so-called flashlight effect is worsened. Traditional cellular
TDD systems have nearby cells with aligned DL/UL
switching points to avoid this matter. We are currently
investigating novel solutions aiming at preserving the UL/DL
flexibility while achieving some degree of predictability of the
interference patterns.
Finally, while IRC provides a strong barrier against
interference and HARQ represents a second tier of protection,
there are still cases in which further interference management
such as ICIC is needed. In the case of the data plane, it is
preferable to have a reactive mechanism which will improve
SINR only when needed and which can respond with certain
agility to the traffic variations. One example is described in
[16], where essentially a dynamic frequency reuse is achieved
in a distributed way. The only requirement for the method in
[16] is that each receiver needs to feedback the post-processing
SINR for the transmitter and some interference dependent
measurement (e.g., Channel Quality Indicator) for all PRBs.
1
0.9
FEASIBILITY OF NETWORK SYNCHRONIZATION
While link synchronization between AP and UE is needed
for coherent data demodulation, network synchronization
among multiple APs is the underlying assumption of our
design since it allows coordinated operations between APs, e.g.
for frequency coordination and interference suppression. In
OFDM-based systems, network nodes are considered
synchronized in case their time misalignment is within a
fraction of the CP duration ( ܶ஼௉ ). In particular, the time
misalignment ߬ெ should fulfill the following requirement:
߬ெ ൏ ܶ஼௉ െ ߬஽ െ ߬ுௐ െ ʹ߬௉
where ߬஽ is the delay spread of the channel, ߬௉ the propagation
delay and ߬ுௐ is the delay response of the hardware filters.
Note that in LTE/LTE-A the propagation delay is compensated
by using a timing advance technique at the UE, which requires
a further hand-shaking procedure before UL transmission can
take place [7]. However, given the lower expected propagation
delay in local area, we believe it is worth to embed it in the CP
at the expense of an extra overhead rather than accepting such
additional latency in the UL.
In a local area scenario, root mean square delay spread and
propagation delay (assuming a 100 m cell radius) are in the
order of ~100 ns and ~170 ns, respectively [17], while the
hardware filter response delay is in the order of ~50 ns. As a
consequence, by assuming the numerology envisioned in Table
I we obtain the following requirement for the time
misalignment:
߬ெ ൏ ͷͳͲ݊‫ݏ‬
Such accuracy level can hardly be achieved with network
solutions such as IEEE Precision Time Protocol (PTP) [18]. In
LTE-TDD, multiple base stations (BSs) can synchronize to the
common reference given by the Global Positioning System
(GPS) satellites. However, penetration losses in indoor make
not possible to rely on GPS signals for network
synchronization. Distributed solutions, where the APs agree
on a common timeline without any centralized coordination,
have then to be pursued for 5G.
Synchronization at network level can be achieved by ensuring
beacon messages exchange among multiple APs, e.g. through
the usage of an opportunistic channel. Such channel can be
mapped, for instance, on the last OFDM symbol in the frame
and used on a contention based manner; the design of the
inter-AP communication channel is for further studies. Each
AP corrects its local timing upon reception of one or more
beacons sent by neighboring APs. In a previous work [19], we
have studied different clock update mechanisms for achieving
tight OFDM synchronization in a distributed manner in a large
network of cells. We evaluated their performance in a local
area 3GPP-inspired scenario with 40 cells of apartment [13],
by assuming different deployment ratios (DRs), i.e. different
probabilities of having an AP at each apartment. We further
assumed clocks having nominal precision of 1 part-per-million
(PPM) and a 10 ms periodicity of the inter-AP communication
channels. Figure 4 shows the CDF of the residual time
misalignment between different APs; in 90% of the cases it is
0.8
0.7
0.6
CDF
IV.
0.5
0.4
DR=25%
DR=50%
DR=75%
DR=100%
0.3
0.2
0.1
0
0
0.05
0.1
0.15
0.2
0.25
0.3
Residual time misalignment [µs]
Figure 4. Residual
synchronization.
time
misalignment
of
distributed
possible to achieve a residual error below 200 ns even for
DR=100%, and then significantly lower than our requirement.
Further details are included in [19]. This justifies our intention
of pursuing distributed synchronization for 5G networks.
V.
NETWORKING ASPECTS
In the previous sections we have introduced our vision
about the lower layers of the 5G RAT. Nevertheless, a more
global picture about the architecture that stands behind that, is
needed in order to understand how this new system is going to
be plugged into the Internet and to the existing mobile
networks.
New service paradigms are appearing fast, expanding the
current Content Delivery Networks. These paradigms are
based on multiple access gateways (GWs) and IP addresses for
hosts, such as “multi-homing” service provision. Besides to
the traditional networks, Internet of Things (IoT) is boosting
the number of connected devices, requiring then the
introduction of novel and dynamic network topologies. All the
motivations will make IPv6 the standard network protocol in
2020.
Furthermore, local area networks like home- and enterprise
networks often utilize shared link IP model for sharing the
same subnet prefix with multiple host and to enable shared onlink services like, e.g. printers, file servers, home
entertainment services.
A 5G system shall naturally support all of it in a smooth way,
integrating itself in different types of network topologies,
possibly based on heterogeneous systems. For all these
reasons, in our vision, a 5G system should be based on an
“Ethernet-over-Radio” (EoR) Link Layer (LL).
Figure 5 illustrates the envisioned 5G network based on the
switched connectivity, integrated in a heterogeneous, multihoming IPv6 network.
Such a design has several advantages in supporting the future
services. Firstly, an Ethernet-type of LL can be easily
translated into an Ethernet IEEE 802.3 LL, allowing the 5G
network to be plugged directly into any existing network,
preserving large flexibility in accommodating the diversity of
traffic applications.
MNO’s Internet GW
Service &
Content Cloud
MNO’s Service GW
REFERENCES
[1]
LocalAR
On-Link
Servers
[2]
[3]
5G AP
[4]
IPv6 link
Figure 5. Cellular access with Ethernet-like, multi homing IPv6based access link.
[5]
[6]
either local area network like home network or the transport
network of a Mobile Network Operator (MNO).
Secondly, it would enable mobile Operating Systems (OSs)
developers not to care about the used access technology when
developing apps or drivers for mobile devices, since many
OSs wish to abstract network interfaces as an IEEE 802 type
interface by default. Finally, EoR will allow cheaper
implementation compared to full IP routing, due the lower
processing power required for operation. But the savings will
come also by the smooth integration with the Software
Defined Network concept, enabling quick, remote, and
eventually automated, network maintenance procedures, e.g.
like those allowed by OpenFlow [20] technology.
VI.
CONCLUSIONS AND FUTURE WORK
In this paper, we have presented our clean slate approach
for a concept design of a novel 5G small cell optimized radio
access technology, which is envisioned as a TDD-based
system. We have proposed a new frame structure which
enables low latency, reduced complexity and power
consumption, as well as native support for interference
coordination schemes and advanced transceivers. A tentative
numerology has been presented.
The flexibility in the UL/DL allocation offered by the frame
structure allows coping with an unprecedented burstiness
level. The usage of inter-cell interference coordination
techniques is foreseen to be beneficial for the control plane,
while the data plane can mostly rely on the native support for
multiantenna IRC receivers enabled by our design as well as
on HARQ. The potential of IRC in boosting the data rate
performance in severe interference limited scenarios has been
discussed, as well as the feasibility of tight distributed network
synchronization which is the main underlying assumption for
our optimized frame based TDD design.
Finally, the usage of an Ethernet-over-Radio link layer is
proposed for an optimal integration with heterogeneous IPv6
networks.
Future work will address the detailed design of novel RRM
solutions (e.g., for scheduling/rank/link adaptation) aiming at
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
B. Raaf et al., “Vision for Beyond 4G Broadband Radio Systems”, IEEE
Proceedings PIMRC, September 2011.
H. Holma, A. Toskala, “LTE Advanced: 3GPP Solution for IMTAdvanced”, Wiley, September 2012.
P. Mogensen et al., “Beyond 4G local area: high level requirements and
system design,” in the Proc. Globecom 2012, International workshop on
emerging techniques for LTE-Advanced and Beyond 4G.
S. G. Larew et al., ”Air Interface Design and Ray Tracing for 5G
Millimiter Waves Communications”, in the Proc. Globecom 2013,
International Workshop on emerging techniques for LTE-Advanced and
Beyond 4G.
H. Koh, C. L. Magee, “A functional approach for studying technological
progress: application toinformation technology “, Technological
Forecasting and Social Change, vol. 73, no. 9, 2006.
D. Gesbert et al. “From theory to practice: an overview of MIMO spacetime coded wireless systems”, IEEE Journal on Selected Areas in
Communications, vol. 49, no. 4, April 2003.
H. Holma, A. Toskala, “LTE for UMTS: OFDMA and SC-FDMA
Based Radio Access”. Wiley, 2009.
M. Lampinen, F. Del Carpio, T. Kuosmanen, T. Koivisto, M. Enescu,
”System-level modeling evaluation of interference suppression receivers
in LTE system”, Proceedings VTC, May 2012.
P. Roshan, J. Leary “802.11 Wireless LAN Fundamentals”, Cisco Press,
2004.
L. G. U. Garcia, I. Z. Kovacs, K. I. Pedersen, G. W. O. Costa, P.
Mogensen, “Autonomous component carrier selection for 4G femtocells
– A fresh look at an Old Problem”, IEEE Journal on Selected Areas in
Communications, vol. 30, no. 3, April 2012.
S. Haykin, M. Sellathurai, Y. de Jong, and T. Willink, “Turbo-MIMO
for wireless communications,” IEEE Communications Magazine, vol.
42, no. 10, October 2004.
F. M. L. Tavares, G. Berardinelli, N H. Mahmood, T.B. Sørensen, P.
Mogensen, ”On the potential of Interference Rejection Combining in
B4G Networks”, accepted in VTC Fall 2013.
“Simulation assumptions and parameters for FDD HeNB RF
requirements”, 3GPP TSG RAN WG4, Tech. Rep. R4-091422, March
2009.
J. Choi, “Optimal Combining and Detection: Statistical Signal
Processing for Communications”. Cambridge University Press, 2010.
T. Yoshino, Y. Sugawara, Y., K. Inagami, J. Tamatsukuri, M. Inaba, K.
Hiraki, "Performance optimization of TCP/IP over 10 Gigabit Ethernet
by precise instrumentation," International Conference for High
Performance Computing, Networking, Storage and Analysis, 2008. SC
2008, vol., no., pp.1,12, 15-21 Nov. 2008
G. W. O. Costa, A. F. Cattoni, I. Z. Kovacs, P. E. Mogensen, "A fully
distributed method for dynamic spectrum sharing in femtocells,"
Wireless Communications and Networking Conference Workshops
(WCNCW), 2012 IEEE , vol., no., pp.87-92, 1-1 April 2012
H. Hashemi, “The Indoor Radio Propagation Channel”, Proceedings of
the IEEE, vol.81, no.7, July 1993.
P.Ohly et al., “Hardware assisted Precision Time Protocol. Design and
Case Study”, 9th LCI International Conference on High Performance
Clustered Computing, 2009.
G. Berardinelli, F. M. L. Tavares, N. H. Mahmood, O. Tonelli, A.F.
Cattoni, T. B. Sørensen, P. Mogensen, ”Distributed Synchronization for
Beyond 4G Indoor Femtocells”, Proceedings of 20th International
Conference on Telecommunications, May 2013.
N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson, J.
Rexford, S. Shenker, J. Turner, “OpenFlow: Enabling Innovation in
Campus Networks,” White Paper, 2008, available online:
http://www.openflow.org/documents/openflow-wp-latest.pd